Using Cellular Automata to Simulate Wildfire Propagation and To
Total Page:16
File Type:pdf, Size:1020Kb
Nat. Hazards Earth Syst. Sci. Discuss., https://doi.org/10.5194/nhess-2018-227 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Discussion started: 6 November 2018 c Author(s) 2018. CC BY 4.0 License. Using cellular automata to simulate wildfire propagation and to assist in fire prevention and fighting Joana G. Freire1 and Carlos C. DaCamara1 1Instituto Dom Luiz (IDL), Faculdade de Ciências, Universidade de Lisboa, 1749-016 Lisboa, Portugal Correspondence: Joana Freire ([email protected]) Abstract. Cellular Automata have been successfully applied to simulate the propagation of wildfires with the aim of assisting fire managers in defining fire suppression tactics and in designing fire risk management policies. We present a Cellular Au- tomata designed to simulate a severe wildfire episode that took place in Algarve (southern Portugal) in July 2012. During the episode almost 25 thousand hectares burned and there was an explosive stage between 25 and 33h after the onset. Results ob- 5 tained show that the explosive stage is adequately modeled when introducing a non-local propagation rule where fire is allowed to spread to the nearest and next nearest cells depending on wind speed. When the rule is introduced deviations in modeled time of burning from estimated time based on hotspots detected from satellite have a root mean square difference of 8.7 hours for a simulation period of 46h (less than 20%). The simulated pattern of probabilities of burning as estimated from an ensemble of 100 simulations show a marked decrease out of the limits of the observed scar, indicating that the model represents an added 10 value for fire fighting in what respects to the choice of locations to allocate resources for fire combat. 1 Introduction Wildfires in the Mediterranean region have severe damaging effects that are mainly caused by large fire events (Amraoui et al., 2013, 2015). Restricting to Portugal, wildfires have burned over 1.4 million hectares in the last decade (Sá and Pereira, 2011), and the recent tragic events caused by the megafires of June and October 2017 have left a deep mark at the political, social, 15 economic and environmental levels. Given the increasing trend in both extent and severity of wildfires (Pereira et al., 2005; Trigo et al., 2005; Pereira et al., 2013; DaCamara et al., 2014; Panisset et al., 2017), the availability of modeling tools of fire episodes is of crucial importance. Two main types of models are generally available, the so-called deterministic and stochastic models. Deterministic models attempt a physics-based description of fires, fuel and atmosphere as multiphase continua pre- scribing mass, momentum and energy conservation, which typically leads to systems of coupled partial differential equations 20 to be solved numerically on a grid. Simplified descriptions, such as FARSITE (Finney, 2004) neglect the interaction with the atmosphere and propagation of the fire front is made using wave techniques. Cellular Automata (CA) are one of the most im- portant stochastic models (Sullivan, 2009); space is discretized into cells, and physical quantities take on a finite set of values at each cell. Cells evolve in discrete time according to a set of transition rules, and the states of the neighboring cells. CA models for wildfire simulation fill a gap between deterministic and empirical models (Rothermel, 1972, 1983), as they 25 prescribe local, microscopic interactions typically defined on a square (Clarke et al., 1994) or hexagonal (Trunfio, 2004) 1 Nat. Hazards Earth Syst. Sci. Discuss., https://doi.org/10.5194/nhess-2018-227 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Discussion started: 6 November 2018 c Author(s) 2018. CC BY 4.0 License. grid. The complex macroscopic fire spread dynamics is simulated as a stochastic process, where the propagation of the fire front to neighboring cells is modelled via a probabilistic approach. CA models directly incorporate spatial heterogeneity in topography, fuel characteristics and meteorological conditions, and they can easily accommodate any empirical or theoretical fire propagation mechanism, even complex ones (Collin et al., 2011). CA models can also be coupled with existing forest fire 5 models to ensure better time accuracy of forest fire spread (Rui et al., 2018). More elaborated CA models that overcome typical constraints imposed by the lattice (Trunfio et al., 2011; Ghisu et al., 2015) perform comparably to deterministic models such as FARSITE, however at a higher computational cost. In the present work, we set up a simple and fast CA model designed to simulate wildfires in Portugal. As benchmark, we have chosen the CA model developed by Alexandridis et al. (2008, 2011) that presents the advantage of having been successfully 10 applied to other Mediterranean ecosystems, namely to the propagation of historical fires in Greece to simulate fire suppression tactics and to design and implement fire risk management policies. This model further offers the possibility to run a very high number of simulations in a short amount of time, and is easily modified by implementing additional variables and different rules for the evolution of the fire spread. We then present and discuss the application of the CA model to the “Tavira wildfire” episode in which approximately 24,800 15 ha were burned in Algarve, a province located at the southern coast of Portugal. The event took place in summer 2012, between July 18 and 21, and fire spread in the municipalities of Tavira and São Brás de Alportel. The Tavira wildfire was one of the largest fires in recent years (excluding the megaevents of the last fire season of 2017), and most of the variables (e.g. total burned area, time to extinction) are well documented and available from official authorities. This fire event was also studied by Pinto et al. (2016), providing a suitable setup for testing the CA model. In addition, comparing the simulation results to 20 this baseline scenario allowed us to identify and formulate the most promising model modifications and refinements to be incorporated in the simulation algorithm. This paper is organized as follows. A description of the fire event to be modeled and of all data required for simulation and validation of results is provided in Section 2. The rationale behind the setting up of the Cellular Automata is described in Section 3. Results are presented and discussed in Section 4, paying special attention to the modeled temporal and spatial 25 deviations from results derived from location and time of detection of hotspots as identified from remote sensing. Conclusions are drawn in Section 5. 2 Data description and processing 2.1 The fire event of Tavira As mentioned in the introduction, we apply a CA model to a large and well documented wildfire that occurred in July 2012 30 in the Tavira and São Brás de Alportel municipalities, located in Algarve, Portugal (Figure 1). The fire was first reported on July 18 (at about 13h UTC) and was considered as contained on July 21 (at about 17h UTC). The fire burned approximately 24,800 ha, mainly shrublands which made up about 64% of the affected area, and spread in heterogeneous, predominantly 2 Nat. Hazards Earth Syst. Sci. Discuss., https://doi.org/10.5194/nhess-2018-227 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Discussion started: 6 November 2018 c Author(s) 2018. CC BY 4.0 License. Figure 1. Left panel: map of Portugal with the location of the Tavira wildfire, where orange represents the burned scar and the black box indicates the study area used in the simulations. Right panels: schematic representation of Europe with Portugal highlighted in blue (top panel) and zoom of the study area (bottom panel). steep terrain. It was the largest wildfire in Portugal in 2012, contributing to more than 22% of the total amount of 110,232 ha of burned area (ICNF, 2012) in that year. The terrain is prevailingly steep, with slopes of 20% located in the higher altitude region in the northern part of the Tavira municipality with hilltops reaching up to 541 meters. The altitude and slope decrease towards the southeast area of Tavira and 5 southwest area of the São Brás de Alportel municipalities, having slopes between 0 and 20% and lower altitudes reaching sea level at several locations (Viegas et al., 2012). Since 2012 was a year of extreme drought, the meteorological background conditions were very prone to the occurrence of large fire events (Trigo et al., 2013). The region of Tavira is characterized by Mediterranean climate, the maximum monthly temperature in August ranging from 25oC to more than 30oC, with maximum absolute temperatures around 39oC, and mean 10 relative humidity below 65% (ANPC, 2012). In 2012, the precipitation in Tavira was 45% below the normal record and the study area had a soil water content value below 10% at the time of the fire (Viegas et al., 2012). The wildfire propensity was further aggravated by above average precipitation in 2010 and 2011, that favored vegetation growth and fuel build up. 3 Nat. Hazards Earth Syst. Sci. Discuss., https://doi.org/10.5194/nhess-2018-227 Manuscript under review for journal Nat. Hazards Earth Syst. Sci. Discussion started: 6 November 2018 c Author(s) 2018. CC BY 4.0 License. Consequently, fire danger as measured by the Canadian Fire Weather Index (FWI) System (Wagner, 1974, 1987; Pinto et al., 2018) was rated Extreme with FWI reaching 56.7 during the fire period (Viegas et al., 2012). The fire propagated in two distinct phases. In the first stage, from 13:00 UTC on July 18 to 17:00 UTC on July 19, the fire burned about 5,000 ha, representing one fifth of the total burned area.